Playing Devils Advocate: I think most people have misunderstood QE's effect on yields.

One of the more common misunderstandings I come across is the idea that the Fed is keeping interest rates artificially low on government securities. This often manifests itself along the lines of:

“The Fed can’t stop QE or yields will shoot up and the government won’t be able to fund the deficit”

or…

“The Fed is funding the deficit. Who will buy the bonds when QE ends?”

and…

“Quantitative Easing increases the demand for bonds, and so yields fall”

People are more than happy to accept that QE drives up equity valuations as investors re-balance portfolios towards risky assets, yet struggle to make the connection that the flip side of this is lower speculative demand for long term bonds and that yields are determined by the market (as things stand currently), so expansionary policy must be reflected in yields. If the market does not price in expansionary policy, it is not functioning correctly, but it does and yields rise under QE. For this reason it is not easy to target yields and inflation simultaneously in free markets. In the current environment yield targeting takes a lower degree of importance to inflation and employment targeting.

This misunderstanding continues for many reasons:

1. A lot of confusion between real and nominal rates, and high yield vs. government debt yields.
2. Trying to fit data to theories rather than the other way around.
3. Because the truth, like many things in economics and finance, is actually very counter-intuitive and hard to accept without deeper thought.
4. Economists are not always traders, and there is some trouble understanding how market participants act.
5. The Fed is an easy target for venting frustrations.

I want to explain why this is misguided, because it leads to a lot of unnecessary fear regarding government funding, as well as poor decision making for unsophisticated investors. It is also leads to a trap where economists call for more QE to keep interest rates from rising, despite a strong possibility that QE is exactly what is making them rise in the first place.

Below I have highlighted the periods in which the Fed was making unsterilized purchases of longer dated Treasuries. If you showed this chart to a child, they would be able to tell you instantly that QE must raise yields, yet somehow many adults are seeing something different. I will discuss why they come to this erroneous conclusion, and why the reality is different to theory. Since I want to discuss the direct effects of long term bond purchases on those yields, I have removed Operation Twist (See discussion below) and purchases of MBS.

Analysis of Data:

Changes in the 10 year treasury yield since the onset of government bond purchases are as follows.

• Periods of balance sheet expansion:
  • QE1: +90 basis points
  • QE2: +51 basis points
  • QE3: +108 basis points
  • Cumulative effect on yields: +249 basis points or +2.49% on the 10 year.
• Periods of no balance sheet expansion:
  • Post-QE1: -82 basis points
  • Post-QE2: -156 basis points
  • Cumulative effect on yields: -238 basis points or -2.38% on the 10 year.

Notice that the 10 year yield is approximately back to its value at the start of the first treasury purchase program. All falls in yields have been associated with the Fed not expanding its balance sheet. All rises in yields have been associated with the Fed expanding its balance sheet. There is no evidence that balance sheet expansion (QE in its current form) reduces government bond yields in the US, and every indication that the opposite is true. If someone can work the data to say that yields have fallen under QE, they surely have their computer screens upside down.

The 10yr Yield and the SP500 is also becoming more positively correlated during each easing program (Yields and the index rise in tandem), with values of 0.55, 0.59 and 0.79 for QE 1, QE2 and QE3 respectively. This is what would be expected from a program that reduces speculative demand for risk free assets and increases speculative demand for riskier assets.

People should not expect to see significantly lower yields at the same time as the index moves higher – which highlights the point of this post: Many see the end of QE resulting in the stock market crashing at the same time as yields rising sharply, resulting in a government that can’t make ends meet. I don’t see the basis for this. The two variables are rising together, and will likely fall together when QE ends and deflationary threats become more pronounced.

Why QE doesn’t increase net demand for Treasuries:

At first glance, there doesn’t appear to be anything wrong with the quote at the top of the post, that QE increases the demand for bonds. To say the opposite – that QE lowers the demand for bonds – sounds insane, because obviously if the Fed purchases 45 Billion of bonds a month they must be increasing the demand for what they are buying. This is incorrect however.

For the Fed to buy 45 Billion per month, the private sector (mostly the household sector, including hedge funds) must be willing to sell 45 Billion per month. The simplest way to understand this counter-intuitive truth is to rephrase the statement:

“The private sector is selling 45 Billion worth of a bonds each month, this will increase supply and yields will rise.”

Because, for every buyer, there is an equivalent seller in each transaction - if the Fed wants to buy, someone must want to sell. It is possible that the percieved benefits of rebalancing towards equities might actually induce greater supply than the Fed's uptake, leading to rising yields. In the case of QE programs, the data shows that speculative sellers (largely households/hedge funds) are the parties selling to the Fed during QE. If QE induces a portfolio rebalancing effect, then it must also increase supply from these parties as they seek to move in to riskier assets.

“Recall, from the analysis above, households are the largest seller to the Federal Reserve for both Treasury securities and MBS. Now we find that purchases of Treasury securities by the Federal Reserve induce households to shift these asset holdings toward corporate bonds, commercial paper, municipal debt and loans, and bank deposits; MBS purchases drive all of the same substitutions, except for bank deposits. In addition, when pension funds sell MBS to the Federal Reserve, they thenshift their portfolio toward repurchase agreements, or very short-term assets. This evidence of shifting investors out of safe assets into riskier assets points to a credible channel for the effects of asset purchases on broader financial markets..." 

"... We find that not all investor types sell to the Fed uniformly. Households (the group that includes hedge funds), broker-dealers, and insurance companies appear to be the largest sellers of Treasury securities when the Federal Reserve buys these securities. Households, investment companies, and to a lesser extent, pension funds, are the largest sellers of MBS when the Federal Reserve buys." 

(Source: From Whom Does the Fed Buy?)

It should be clear from this understanding that the quantity purchased does not matter if the private sector sells willingly, and that prices will not be driven either way by this process alone. The Fed’s purchases do reduce the total outstanding supply in the market by moving the bonds on to the Fed’s balance sheet until maturity, relative to the counter-factual where the Fed made no purchases – this does not result in lower yields because QE simultaneously reduces new private sector demand for safe assets in favour of risky assets.

Taking this portfolio effect further, it also needs to be emphasized again that these are willing sellers. The Fed is not forcing these parties to relinquish their bonds, they sell because they see better opportunities elsewhere. There is no reason that the Fed's purchases would add to existing demand, because those that sold to the Fed will not reinvest their freshly issued deposit back into those same treasuries - because for them to do so would be a meaningless transaction. There is no incentive or reason for any investor to sell a bond to the Fed and then buy an equivalent bond with the proceeds.

It might be the case that sellers of long term treasuries may sell in order to purchase shorter duration bonds - this is probably occuring now as the Fed strengthens forward guidance regarding the Fed funds rate. Short rates are expected to stay at zero for longer, even if the Fed tapers its long term bond purchases, a message Bernanke is making sure to emphasize as tapering becomes more likely. This will put pressure on long term interest rates.

A simple diagram showing hypothetical effects – the reality is more likely that private sector demand falls by more (hence the rising yields) and Treasury keeps issuing new securities, however I just want to demonstrate how the portfolio effects of QE can net out to roughly zero and should not drive yields lower.

Furthermore, the Fed is a price taker in these transactions, not a price maker. That is, the Fed conducts a series of competitive auctions with its Primary Dealers (1). In this process, dealers submit offers to the Fed’s dealing desk and the Fed accepts the lowest priced offers received to fill it’s purchase amount (2). The Fed’s purchases in no way drive yields lower, because the Fed must accept voluntary offers from the private sector and the market comes to a price based on a number of other factors.

Why might QE raise yields?

With the exception of Operation Twist, QE has unambiguously been associated with rising yields and the end of QE programs have always resulted in this process reversing – leading to plummeting yields. The same occurred in Japan when Abenomics was supposed to drive long term rates lower, but instead yields on the 10 year JGB doubled under the program. Abe and Kuroda were surprised by this, but they should not have been after seeing the same effect played out many times before.

Firstly, when the Fed creates reserves ex nihilo, the market views this as expansionary and to some degree inflationary and prices this in to yields accordingly. Fed watchers over the last couple of years understand that banks don’t lend reserves (except to each other) and they are not reserve constrained. The aggregate of market participants however has a knee-jerk reaction to balance sheet expansion – that it is inflationary. From my understanding there is no direct effect on inflation, but through various second order effects such as exchange rate depreciation and lowering yields on risky credit instruments, there is probably some minor inflationary effects (though we can never know for sure without the counter-factual of no QE at all).

Because Operation Twist a) Was sterilized, and did not involve balance sheet expansion, and b) Targeted yields directly, it was not seen as particularly expansionary and inflationary by the market. Beyond this, traders will not fight the Fed if it wants to target yields. As quantitative easing is currently constructed, it targets a quantity of purchases but has no set yield target. This means firstly that the Fed does not have any control over yields, and that traders are not fighting the Fed by selling into the Fed’s purchases.

Secondly and more simply, markets ‘buy the rumour and sell the news’. This can mean pricing the event into the market in the lead up to announcement, or pricing it out of the market – dependent on what the market expects.

In the case of QE programs, the market front runs the Fed’s purchases by buying heavily before the programs begin, then ‘takes profit’ when the Fed enters the market, leading to selling pressure on treasuries and rising yields rather than falling yields. Purely academic economists sometimes struggle with this, but event driven traders will know that if an event such as tapering is to be announced, traders will continue to push yields higher until it is actually announced, then flood back into the market on the actual announcement, perhaps in anticipation of the next program – when they will again sell to the Fed. Rinse and repeat.

Finally, portfolios re-balance. QE makes treasuries unattractive and equities and high yield instruments more attractive. People are simply not willing to hold treasuries at the same low interest rates while QE is supporting risk assets, and they adjust their pricing accordingly.

QE can lower yields in at least 4 scenarios:

1. If the purchased quantity is too low to offset deflationary fears.
2. If the purchase program targets yields directly rather than a set quantity of assets.
3. If the program is sterilized and there is no central bank balance sheet expansion.
4. If yields are rising because there is sovereign default risk (Euro Area, sub-optimal currency area), QE can lower the risk premium by guaranteeing the solvency of the country.

However the Fed only has the ability to target one variable at a time. Richard Koo has explained this a few times, however I will go over it also:

1. If the Fed sets a firm inflation target, it loses control over what quantity of assets it must purchase to achieve this and cannot control yields (as the market will price in the inflation and yields rise).
2. If the Fed sets a monthly quantity to purchase, it controls it’s purchase amount, but can never really force a set inflation rate or maintain control over interest rates (hence the Fed hasn’t been able to achieve it’s target inflation rate or keep yields down).
3. If the Fed targets yields, then it cannot control the amount of assets it purchases to achieve and defend the yield and cannot control the rate of inflation. It might actually be the case that the market respects the Fed’s target yield, and so the Fed ends up buying very little from the market to defend its target – in this case inflation would be hard to come by. The opposite could be true in a central bank that has less credibility – the central bank might be forced to buy vast quantities of assets to control yields – in such a case the country might find itself in trouble.

Why make the claim that QE lowers treasury yields?

This is really hard to figure out, since it seems so clear from the evidence and the logic that QE hasn’t done anything to bring yields down. I have some suspicions.

1. Some economists set out to prove a point, then find data to support their perspective. I haven’t read every paper and article written on QE’s effects, but from the ones I have that claim a reduction of yields through QE I notice two ways in which the methodology seems disingenuous.
   1. By selecting arbitrary dates between which to run a regression instead of simply the period in which the program was actually running – this includes dates of announcements and discussions. If they regressed only the period between which the purchases took place I cannot imagine how they could come to the conclusion that yields moved lower under QE.
   2. By examining the whole period since the onset of the crisis. This ignores the deflationary effect of a bursting credit bubble (debt deflation) over the period in which yields kept moving lower, only to be periodically raised by QE programs. It also ignores the 30 year trend of yields moving lower due to a glut of savings and low inflation.
2. People confuse real rates with nominal rates. Some discussion argues that QE lowers the real interest rate by inducing inflation. This is obviously not the same thing as the nominal interest rate and I do not dispute that QE can lower the real interest rate. The problem is that this distinction is not made clear enough in much of the financial media, and so the point gets muddied and muddled until it is assumed that QE lowers nominal rates.
3. People confuse Treasury yields and the effect on high yield and corporate debt markets. QE has helped lower funding costs here, so QE does lower yields – just not for risk-free assets like treasuries. This is another point that gets confused and blended together, leading to incorrect conclusions.

Conclusions:

Though it can never be known for sure, there is a strong case to be made that the Fed is keeping rates artificially high, rather than low. Trying to picture the counter-factual – where the Fed did not begin its QE programs, it seems like a no-brainer that a debt deflation scenario, with an oversupply of savings would lead to some very low yields. Demand for credit would be incredibly low.

It is incorrect that the US Treasury will have difficulty funding itself when QE ends. I will need to explain this further in another post, but from a purely mechanical point of view QE neither lowers the nominal rate on bonds nor is it direct funding (as the Fed purchases in the secondary market). To the extent that QE may help to lower real yields, this may help reduce the significance of the debt over a longer term. From a short term perspective however, nominal yields will be lower without a QE program running so at best the effect is ambiguous as to whether it would be easier or harder for Treasury to fund.

For short term interest rates such as the Fed Funds rate, the Fed also is forced to keep rates higher than they otherwise would be (3) – this occurs because of the large quantity of excess reserves that accumulate due to banks inter-mediating purchases from the household sector. The Fed needs to pay interest on these reserves to maintain control over the Fed Funds rate, which, in the case of large excess reserve balances and no interest paid would see the Fed Funds fall to zero as banks sought to loan their excess reserves. By putting a floor under rates the Fed maintains control over its policy rate.

A Final Thought:

Explicit in my thinking here is that while QE is running, yields and the indices are positively correlated. This poses an interesting question – if QE makes risk free yields rise, and conversely makes yields on risky assets fall, will there come a point where the psychology behind QE ‘flips’ and stops working because the yield on risk free assets has risen enough that they become attractive again?

My belief is that, yes, this is the ultimate outcome if QE continues running indefinitely. The portfolio re-balancing of QE is largely psychological and not mechanical, and it can change if valuations get too far out of line or sentiment too bullish on equities and risk assets. At what point this would happen, I am not sure, but I think the idea that multiple expansion will continue regardless of anything else until QE ends is not correct (though it may continue for a while). When relative value differentials become large enough, Treasuries will look attractive and equities repulsive – especially considering the lack of inflation achieved by QE.

1. FOMC: Statement Regarding Purchases of Treasury Securities.

2. POMO FAQ.

3. Why pay Interest on Reserves?